PROJECT SUMMARY Schizophrenia is an often devastating neuropsychiatric illness. Genetic factors have been strongly implicated. The genetic contribution to schizophrenia is, however, likely to be complex and difficult to capture by merely considering sequence variation. DNA methylation studies represent a particularly promising complement for several reasons. First, as methylation is directly related to gene expression, knowledge of gene methylation levels may add to the prediction of disease status. Second, methylation studies can provide insight into phenomena such as age of onset, the episodic nature of schizophrenia, gene-environment interactions, parental effects, and sex differences. Third, methylation sites are also excellent new drug targets as they are modifiable by pharmacological interventions. Finally, methylation markers are accessible at the stable DNA level, which from a translational perspective means that they can potentially also be used in clinical settings to improve diagnosis and treatment. Methylation studies have historically been restricted to a limited number of candidate genes. However, it has recently become technically and economically feasible to measure the methylation status of millions of markers simultaneously. This resembles recent developments in genomewide association studies (GWAS), which have accelerated the discovery of disease variants considerably. To identify methylation sites related to schizophrenia, we have applied statistical theory developed by our group to determine the most cost-effective study design. Based on these optimal design calculations, we propose a whole genome methylation profiling study in 750 schizophrenia cases and 750 controls. To eliminate false discoveries due to technical and sampling errors, we will follow up the most promising sites in an independent sample of schizophrenia cases and controls using pyrosequencing. Our current power calculations suggest that we may need to follow up 65 regions in 800 cases and 800 controls. However, one strength of the statistical method we developed is that it can be used adaptively--that is, the optimal sample size and number of markers for the second pyrosequencing replication stage can be empirically determined based on parameters estimated from the first, array-based discovery stage. In order to improve our understanding of the disease mechanisms, secondary analyses will be performed relating the methylation regions to clinical information as well as to a genome-wide panel of already available SNP markers and copy number variant calls. PUBLIC HEALTH RELEVANCE: Schizophrenia is an often devastating neuropsychiatric illness. DNA methylation studies represent a particularly promising complement to current genetic studies that offer great potential to improve the understanding and treatment of the disease.